The Application of AI Tools in the Media
Industry
On October 22nd, we attended a lecture led
by a team leader from Edelman HK, who provided an in-depth introduction to the
development and use of generative AI.
He demonstrated how AI can generate videos
based on text descriptions, noting that while AI-generated videos a year ago
were filled with distortions in shapes and spaces, making them easily
recognizable as artificial, the current generation of videos not only meets
user content demands but also significantly improves in terms of detail and
camera logic, making them much harder to distinguish from real content.
The speaker mentioned an objective
observation: users tend to overestimate the impact of new technologies in the
short term but underestimate their long-term influence. I completely agree.
When ChatGPT first emerged in 2022, numerous media outlets claimed that it
would herald the Fourth Industrial Revolution, predicting that AI would soon
become indispensable to our lives. These articles clearly overestimated the
speed of generative AI’s development. However, as more than a year has passed,
many ordinary internet users feel that generative AI has not significantly
impacted their daily lives. Yet, AI has already taken root in various
professional fields and even changed the operational methods of certain
industries, with most ordinary users underestimating its influence.
When discussing AI usage techniques, the
speaker emphasized that it is crucial to provide AI with personal information
and background so that the AI can generate responses aligned with user input.
The more detailed the question, the higher the quality of the AI’s response.
Conversely, shorter and more general questions lead to less specific answers.
This provided me with valuable insights for using generative AI more
effectively.
He also introduced us to the transformative
updates generative AI tools have brought to traditional media production tools,
such as Adobe Firefly embedded in Adobe Photoshop. This tool allows users to
select specific areas of an image and replace them with AI-generated content
that blends seamlessly into the original background. I have tried this feature
in Photoshop, though Adobe later imposed regional restrictions on its usage.
Regarding the application of generative AI
in the media industry, I have the following insights:
Automation of Content Creation
Content creation is the most critical phase
of content production. In advertising, generative AI can now produce
high-quality promotional images, posters, and even video content that resemble
real-life footage, replacing the need for companies to hire actors for
advertisements. Moreover, AI-generated materials help avoid copyright issues
when using certain assets.
AI’s data processing capabilities excel in
tasks requiring large amounts of data, simple logical connections, and high
accuracy, making it suitable for generating structured reports. Currently, AI
writing is particularly effective for producing standardized, template-based,
and repetitive content, such as financial news, sports updates, and weather
forecasts, primarily in the form of brief news reports and financial
statements. Additionally, AI writing greatly increases the speed and volume of
content production while continuously reducing production costs.
Some platforms are transitioning their
content production models from PGC (Professional Generated Content) and UGC
(User Generated Content) to AGC (Automatic Generated Content).
Intelligent Content Distribution
AI’s breakthroughs in algorithms lie in its
ability to continually improve recommendation systems, balancing broad
relevance with individual user preferences. AI can gather and analyze user
behavior, interests, and location on media platforms, building and refining
user profiles to provide precise, intelligent content recommendations. Compared
to manual content distribution, AI offers two key advantages: a vastly expanded
distribution range and the ability to analyze behavioral data to discover
potentially relevant content beyond users’ primary interests.
Efficient Content Moderation
In the digital era, manually reviewing the
vast amount of content and data produced online is time-consuming and
labor-intensive, while AI holds significant advantages in this area. AI can
review video content by using algorithms and other technologies to analyze and
evaluate the type, elements, and keywords of videos, ensuring compliance with
content regulations. AI technology can also monitor copyright issues by
conducting data mining, image recognition, and data comparison to track content
distribution across platforms and detect unauthorized reproductions and
modifications. It can quickly identify infringements such as unauthorized
reprints, plagiarism, and content theft, generating real-time copyright
monitoring reports. In fact, many video platforms have already implemented AI
for content moderation. For instance, Bilibili initially relied on manual video
review, which could take up to four hours during periods of high submission
volumes due to limited human resources. However, after adopting AI-based review
systems, the number of required reviewers dropped significantly, and the review
time was reduced to under one minute, greatly improving efficiency.
In conclusion, AI has undoubtedly opened up
vast new opportunities for the media industry. As we look to the future, AI is
certain to occupy a central role in the media landscape and may even become
increasingly dominant.
No comments:
Post a Comment